RESEARCHER PROFILE: PhD/ R1: First stage Researcher
RESEARCH FIELD(S)1: Computer Science
MAIN SUB RESEARCH FIELD OR DISCIPLINES1: Medical Science
JOB /OFFER DESCRIPTION
The AI-CARE PhD project aims to design a hybrid and interpretable clinical decision support system (CDSS) for anesthesia, combining symbolic artificial intelligence with statistical learning. In France, more than four million anesthetic procedures are performed yearly, each requiring a preoperative anesthesia consultation. During this critical step, anesthesiologists must synthesize a vast amount of heterogeneous data—most often unstructured and dispersed across electronic medical records—such as clinical notes, medication lists, comorbidities, lab values, and surgical risk factors.
The scientific challenges of this project are significant. First, extracting and structuring relevant clinical information from unstructured texts involves advanced natural language processing (NLP) tailored to medical content. This includes entity recognition (e.g., medications, diagnoses), relation extraction (e.g., comorbidity–treatment links), and concept normalization using terminologies such as CIM-10, ATC, and SNOMED-CT. Second, modeling anesthetic decision-making requires integrating heterogeneous data (structured and unstructured), capturing expert reasoning under uncertainty, and learning risk stratification patterns from large cohorts. Third, embedding formal medical knowledge (e.g., guidelines from SFAR or ESAIC) into a symbolic engine involves building a domain-specific ontology and translating best practices into machine-actionable rules. Lastly, the system must adapt to real-world outcomes via feedback loops, learning from perioperative events to refine its predictions over time.
Phase 1 focuses on the development of an AI pipeline that processes anesthesia consultation notes and produces structured, analyzable data. This will involve adapting pretrained biomedical language models (e.g., ClinicalBERT, CamemBERT médical, LLaMA-Med) to French clinical data and developing an anesthesia-specific ontology to support symbolic reasoning. Using a cohort of over 80,000 anonymized consultations, supervised models will then be trained to predict anesthetic risk (e.g., ASA class) and suggest context-aware management strategies.
Phase 2 will validate and refine these models using a second dataset of 35,000 perioperative records. The goal is to implement adaptive learning: comparing predicted strategies and actual clinical decisions, correlating them with real intraoperative events (hypotension, difficult intubation, adverse drug responses). This phase introduces a feedback loop, allowing the AI system to adjust its recommendations and confidence levels based on real outcomes—thus creating a self-improving, continuously learning CDSS.
This interdisciplinary PhD bridges clinical anesthesiology, data science, and symbolic reasoning. It contributes to the field of explainable and ethical AI in medicine. The thesis will be hosted at the Laboratoire de Biomécanique Appliquée (Faculté de Médecine de Marseille), in collaboration with the Hôpital National d’Instruction des Armées Sainte Anne (Toulon) and APHM - Hôpital de la Timone. It is part of the strategic program "Sciences numériques et IA pour la santé" led by the Institut Laënnec and supports the French roadmap for trusted medical AI.
TYPE OF CONTRACT: TEMPORARY
JOB STATUS: FULL TIME
HOURS PER WEEK 35
APPLICATION DEADLINE: 01/10/2025 09:00 am
ENVISAGED STARTING DATE: 01/10/2025
ENVISAGED DURATION: 36 months
JOB NOT FUNDED THROUGH AN EU RESEARCH FRAMEWORK PROGRAMME
WORK LOCATION(S): Laboratoire de Biomécanique Appliquée, Faculté de Médecine Nord, 51 Bd Pierre Dramard CS80011, 13015 Marseille, France
WHAT WE OFFER:
The PhD fellow will benefit from:
- Access to a large-scale, anonymized clinical dataset
- Dual supervision by experts in anesthesiology and artificial intelligence
- A dynamic, interdisciplinary research environment (LBA-AMU)
- State-of-the-art computing infrastructure
- Networking within the national AI-health ecosystem (Institut Laënnec, Assitance Publique Hôpitaux de Marseille, Aix Marseille Université, French Army Health service)
- Opportunities to valorize the work through scientific publications and clinical applications (e.g., integration into hospital software)
Additional information: The Euraxess Center of Aix-Marseille Université informs foreign visiting professors, researchers, postdoc and PhD candidates about the administrative steps to be undertaken prior to arrival at AMU and the various practical formalities to be completed once in France: visas and entry requirements, insurance, help finding accommodation, support in opening a bank account, etc. More information on AMU EURAXESS Portal
QUALIFICATIONS, REQUIRED RESEARCH FIELDS, REQUIRED EDUCATION LEVEL, PROFESSIONAL SKILLS, OTHER RESEARCH REQUIREMENTS
- Master’s degree (MSc or equivalent) in biomedical engineering, health informatics, computer science, applied mathematics, or a related field
- Knowledge of medical terminology and clinical workflows is a strong asset
- Skills in natural language processing, machine learning, or symbolic AI (or willingness to acquire them)
- Programming experience (Python, SQL, scikit-learn, PyTorch, etc.)
- Familiarity with medical ontologies or interest in semantic technologies is welcome
- Prior experience with clinical data or participation in health-related projects will be considered a plus
Soft skills:
- Scientific curiosity and critical thinking
- Strong analytical and problem-solving abilities
- Ability to work independently and in a multidisciplinary team
- Good communication skills (written and oral)
- Fluency in English; knowledge of French is a plus since the health data are in french
- Adaptability, sense of responsibility, and rigor
REQUESTED DOCUMENTS OF APPLICATION, ELIGIBILITY CRITERIA, SELECTION PROCESS
- Motivation letter
- CV
- Master’s‐level academic transcript (with grades)
- Reference letters (optional but appreciated)
HOW TO APPLY:
